Mapping Seagrass Cover Percentage and Estimation of Aboveground Carbon Stock using PlanetScope SuperDove Imagery and Random Forest Algorithm in Bontang, East Kalimantan Diki Akhyar Amanatulloh (a*)- Pramaditya Wicaksono (b)- Wirastuti Widyatmanti (b)
a) Master of Science in Remote Sensing, Faculty of Geography, Universitas Gadjah Mada, Sleman, Daerah Istimewa Yogyakarta 55281, Indonesia
b) Department of Geographic Information Science, Faculty of Geography, Universitas
Gadjah Mada, Sleman, 55281, Indonesia
Abstract
Seagrass meadows are a vital component of blue carbon ecosystems, playing a significant role in carbon sequestration and storage, and contributing effectively to climate change mitigation. However, the sustainability of seagrass ecosystems is increasingly threatened by climate-induced degradation and anthropogenic disturbances. Therefore, spatial inventory of seagrass meadows including aboveground carbon stock (AGCseagrass) estimation is essential for conservation and climate action. High-resolution remote sensing imagery from PlanetScope SuperDove, integrated with field survey data, offers an optimal approach for mapping benthic habitats, seagrass percent cover (PC), and AGCseagrass. This method is particularly effective and efficient for covering extensive areas with diverse terrain conditions. The aim of this research is to map benthic habitats, seagrass percent cover, and aboveground carbon stock using PlanetScope SuperDove imagery with a spatial resolution of 3 meters. The Random Forest algorithm was employed for both classification and regression, integrating remote sensing data with field observations. The results indicate the presence of five benthic habitat classification classes, with an overall accuracy (OA) of 96.44% and a kappa accuracy of 95.16%. For seagrass percent cover modeling, the Random Forest algorithm yielded an R2 of 0.66 (RMSE = 21.09), and an average seagrass percent cover of 67.06%. The estimated total aboveground carbon stock (AGCseagrass) in Bontang reached 984,915 tons C, with an average value of 274.05 g C/m2, calculated across a spatial extent of 32.35 km2. The AGC modeling also resulted in an R2 of 0.66 (RMSE = 9.25).
